Impact Of Weekly Temperature Variability On Influenza-like Illness And Prediction Early Warning Model In Shandong Province

BackgroundClimate change is an undoubted fact.In the past half century,the global average temperature and temperature variability have shown a general upward trend.Average temperature is an index to measure the average level of temperature,which reflects the exposure level of the general population in the area where the observation point is located.Temperature variability refers to the periodic and regional range of temperature change,which reflects the instability of temperature.For the human body,it is more difficult to adapt to the changing temperature than the stable ambient temperature.Studies show that both short-term and long-term temperature changes may cause damage to health,and affect a variety of body functions,including cardiovascular system and respiratory system.At present,many scholars have done a lot of research on the relationship between average temperature and human health.However,the evidence about the possible impact of temperature variation on health is mostly focused on the impact of temperature variation on non-accidental death,and the research related to the risk of infectious diseases is still less.Influenza is a public health problem that needs to paid close attention because of its rapid onset,rapid transmission and strong virus variability,which makes it difficult to form a stable long-term immunity in the population and is prone to outbreaks and epidemics.China now has an Influenza surveillance network covering 31 provinces,cities and regions,which carries out surveillance on Influenza-like Illness(ILI)cases,and comprehensively reflects the epidemic intensity of Influenza through the percentage of influenza-like cases and the positive detection rate of Influenza cases.The incidence of ILI presents obvious seasonality,and meteorological factors have a certain influence on the incidence of ILI.At present,most relevant studies focus on the effect of temperature and humidity on the incidence of ILI,while few studies involve the effect of temperature change on the risk of influenza or ILI,and the effect in this aspect still needs further exploration.Forecasting and early warning of disease epidemics and outbreaks are helpful for relevant health departments to carry out disease prevention and control work in time.The monitoring level of meteorological factor data is stable,the accessibility is good,and the effect on disease has a certain lag.Using meteorological factors to establish the prediction and early warning model of infectious diseases can provide the decision-making proposal for the prevention and control work of health departments.Objectives1.Describe the characteristics of ILI in Shandong Province from 2016 to 2018.2.To explore the effect of temperature variation on the risk of Ili in Shandong Province at the urban level,regional level,and provincial level.3.To explore the effect of temperature variation on the risk of ILI in different age groups,and to identify vulnerable groups under the influence of temperature variation.4.Identify the temporal and spatial clustering areas of ILI in Shandong Province from 2016 to 2018,and establish the prediction and early warning model of ILI based on meteorological factors.Methods1.Data sourcesThe ILI data of Shandong Province from 2016 to 2018 are from Shandong influenza case monitoring system,and the meteorological data are from the China Meteorological data sharing service network(http://data.cma.cn/)The weekly temperature variability index was calculated by daily minimum temperature and daily maximum temperature.The demographic data of each city are from Shandong statistical yearbook and Shandong 2010 census data.2.Statistical analysis(1)The influenza-like-illness percentage(ILI%)and weekly temperature variability(WTV)were calculated.Descriptive statistical analysis was used to describe the distribution and change trend of ILI in the different time,age groups,and space of the province.(2)The DLNM-meta two-stage model was used to analyze the impact of WTV on the risk of ILI at a city level,regional level and provincial level.(3)The subgroup analysis was carried out to identify the vulnerable population under the effect of WTV(4)The spatiotemporal scanning analysis was used to detect the spatiotemporal distribution of ILI in Shandong Province from 2016 to 2018,to find out the high concentration area of ILI incidence,select a representative city in the high concentration area,and use support vector regression machine to predict and early warn the incidence level of Ili in this area.ResultsFrom 2016 to 2018,sentinel hospitals in Shandong Province reported 1,448,652 cases of ILI surveillance.The total ILI%for three years was 4.94%,and the ILI%for each year was 4.79%,5.72%,and 4.33%,respectively.The incidence of ILI presents obvious seasonality at the provincial level,there are two main peaks in a year,summer,and winter,and the time and peak of the annual peak are slightly different.The number of reported ILI cases and ILI%in Qingdao and Jinan are the top two in the province.Qingdao’s ILI%in 2016 and 2017 were both above 20%,and the ILI%in 2018 was 10.98%,which is 2-5 times of the average level of the whole province.In terms of population distribution,among all age groups,the number of ILI cases in the 0～4 years old group is the highest,with the highest ILI%,and the 15～24 years old group and the age group over 60 years old have the lowest ILI%.Among all reported ILI cases,cases under 15 years old account for the ratio exceeds 86%.During the study period,the average WTV of the province was 5.52,the minimum was 1.9,the maximum was 11.07,and the Inter Quartile Range(IQR)was 2.11Taking the Pas of WTV as a reference,the effect of WTV on ILI is nonlinear at the provincial level,both high WTV and low WTV can increase the risk of ILI.When WTV is lower than the reference value in modeling,the risk of ILI first increases and then decreases with the trend of WTV.When WTV=4.5,the RR reaches the peak of this stage.When WTV is higher than the reference value in modeling,the exposure-response relationship between WTV and ILI is linear.After adjusting the effect of the weekly average temperature on ILI,the effect still exists,and the combined cumulative efifect curve is roughly the same.There are certain differences in the effect of WTV on ILI between different regions.In inland areas,high-temperature variation areas,and areas with a high proportion of people aged 0-14,WTV can increase the risk of ILI under both low and high conditions.In coastal areas,high WTV and low WTV do not increase the risk of ILI.In the low-temperature variation areas,when WTV is at a low level,it has a protective effect on the risk of ILI.In areas with low proportions of a low proportion of people aged 0-14,the impact of WTV on ILI is linear.Vulnerable population analysis shows that there are certain differences in the vulnerable populations with ILI incidence under the effect of temperature variation.The vulnerable populations under low-level of WTV include the age groups of 0～4 years old,25-59 years old,and over 60 years old.The vulnerable groups under high-temperature variation are the 5-14 years old,15-24 years old,and 25-59 years old age group.The temporal and spatial clusters of ILI in Shandong Province are mainly concentrated in central Shandong and northwestern Shandong.The time clusters are between the first week of December 2017 and the second week of February 2018.The support vector regression machine prediction model for the first-type cluster of Dezhou City fits well.The adjusted R2,RMSE,MAPE of ILI%are 0.89,0.39,0.07,respectively.The adjusted R2,RMSE,MAPE of ILI case number are 0.86,35.85,0.12,respectively.The early warning department selected P80(314)of the number of ILI cases and P90(7.33%)of ILI%as the early warning thresholds for the arrival of the peak of the influenza epidemic in Dezhou.The AUC of the early warning model based on ILI%and ILI reached 0.9655 and 0.75.All the indicators showed a good warning effect,this model could be applied in other high-risk areas.Conclusion(1)At the provincial level,the variation of ILI in Shandong Province is seasonal,which is high in winter and summer,and the time of the peak is slightly different each year.The incidence rate of Qingdao is the highest one among all the cities,which subordinated to Shandong Province,and the population of 0-14 years old is the high-risk population.(2)Taking the P75 of WTV as the control value,both higher and lower WTV at the provincial level can increase the risk of ILI.There are some differences in the effect of WTV on ILI in different regions.WTV mainly affects inland areas,high temperature variation areas and areas with a relatively high proportion of people in the 0-14 age group.Vulnerable groups with low WTV are 0-4 years old,25-59 years old,and over 60 years old,and vulnerable groups with high WTV are 5-14 years old,15-24 years old,and 25-59 years old.(3)The time and space gathering area of ILI in Shandong Province is in northwestern Shandong,and the time is concentrated from the first week of December 2017 to the second week of February 2018.(4)ILI forecasting and early warning models based on meteorological factors have good effects and strong operability,and can be further promoted and used.